Solid State Physics

TIGHT BINDING MODEL Lecture 20

A.H. Harker Physics and Astronomy UCL 7.6 The tight-binding model 7.6.1 Overview For materials which are formed from closed-shell or ions, or even covalent solids, the free model seems inappropriate. In the tight-binding model, we imagine how the wavefunctions of atoms or ions will interact as we bring them together.

2 For example, take two hydrogen atoms, A and B, and consider the states ψA ± ψB.

3 The symmetric (+) form has more screening charge between the nu- clei, and has lower energy.

4 When more atoms are brought together, the degeneracies are further split - to form bands ranging from fully bonding to fully antibonding. Different orbitals can lead to band overlap.

5 7.6.2 Tight-binding theory Consider an element with one per unit cell, and suppose that each atom has only one valence orbital, φ(r). Then we can make a wavefunction of Bloch form by forming −1/2 X ψk(r) = N exp(ik.Rm)φ(r − Rm). m Confirm that this is a Bloch function. If T is a translation vector:

−1/2 X ψk(r + T) = N exp(ik.Rm)φ(r − Rm + T) m −1/2 X = N exp(ik.T) exp(ik.(Rm − T)φ(r − (Rm − T)) m = exp(ik.T)ψk(r) because if Rm is a lattice vector, so is Rm − T.

6 Find the expectation energy of the Hamiltonian: −1 X X hk|H|ki = N exp(ik.(Rn − Rm))hφm|H|φni m n where φm = φ(r − Rm). Now hφm|H|φni will be large if n and m are the same atomic site, or nearest neighbours, but will decrease rapidly with separation. Write hφn|H|φni = −α, hφm|H|φni = −γ if n and m are nearest neighbours, hφm|H|φni = 0 otherwise. Then X Ek = hk|H|ki = −α − γ exp(ik.Rn), n where the sum is over nearest neighbours only, and Rn is a vector joining an atom to its nearest neighbours. For example, in two- dimensional square lattice we have {Rn} = {(a, 0), (−a, 0), (0, a), (0, −a)} so that if k = (kx, ky) Ek = −α − 2γ(cos(kxa) + cos(kya)). 7 Clearly, as cos ranges between −1 and 1 Ek ranges between −α − 4γ and −α + 4γ, giving a band width of 8γ. Near k = 0 we can expand the cos functions as 1 cos θ ≈ 1 − θ2, 2 so 1 1 E ≈ −α − 2γ(1 − k2a2 + 1 − k2a2) k 2 x 2 y 2 2 2 = −α − 4γ + γ(kx + ky)a which is free-electron-like, giving circular constant-energy surfaces near the centre of the Brillouin zone. If both kx and ky are close to π/a, write π π k = − δ k = − δ , x a x y a y so that, remembering cos(a − b) = cos(a) cos(b) + sin(a) sin(b),

8 we have

Ek = −α − 2γ(cos(π − δxa) + cos(π − δya)) = −α − 2γ(cos(π) cos(δxa) − sin(π) sin(δxa) + cos(π) cos(δya) − sin(π) sin(δya)) = −α + 2γ(cos(δxa) + cos(δya)) 2 2 2 = −α + 4γ − γ(δx + δy)a giving circular constant-energy surfaces near the zone corners too. Finally, in the middle of the band

cos(kxa) + cos(kya) = 0, the solutions to which are of the form

kxa = π − kya, or straight lines.

9 Finally, then, we have the constant energy surfaces for this tight- binding model.

10 7.6.3 Comments on tight binding theory • Note that band width depends on two-centre integrals (γ): for transition metals, this leads to narrow d-bands and wide s-bands. • Near the top and bottom of bands, we have quadratic dependence on k.

11 A real band structure.

12 13